successionuniversefandomcom-20200215-history
Mest Kölleg Point Test
Origin The Mest Kölleg Point Test was a joint psychological and cybernetics systems test developed by joint Nemean staff and unofficially headed by Dr. Mest Kölleg. Although the test was eventually named after Dr. Mest Kölleg, the resident psychiatrist on staff of the Argo, the test was originally devised by a team of lower level "grunt" scientific assistants and minor research staff once they realized that Nemean members coincidentally all had Ark units with sentient AIs. The purpose of the test was to evaluate these AIs and monitor them, and furthermore to evaluate the possible differences that are created from the development of civilian AIs and military AIs. However early into the test's theoretical creation the creators realized they lacked the rank and prestige to actually institute such a test. Thus they reached out to Dr. Kölleg and implored her to head the team and institute the test. She did, and when the test was created and instituted, the test was named after Dr. Kölleg. Structure In order to take part in the Point Test, an AI is temporarily barred from accessing it's own memory and emotional databanks, leaving it only it's personality and language banks to be accessed. Once this has been done the AI is led/taken into a room that is devoid of human presence. Somewhere in this room is a Point Test "Analyst," a device synthesized by Argo engineering team on request of the Point Test team. The "center" or core of the Analyst where the handle protrudes from is a rudimentary virtual intelligence or "dummy", non-sentient artificial intelligence that serves as the test administrator, who interacts verbally with the test taker and records it's answers. If the test taking AI is still in it's Ark when taking the test, The four camera nodes that are attached to the side of the Analyst closely monitor the Ark's frame for verbal queues such as slight twitching or hand gestures. This is so that the test is delivered in an emotionally neutral environment as possible. Test takers and Ark AIs taking the test are to answer questions as immediately as possible and are instructed not to hesitate on answering questions. Hesitation when answering questions means that an AI might have emotional or memory banks that are corrupt and have "bled" into the personality data-bank. Once the test is completed, the AI is re-introduced to it's memory and emotional banks. The Point Test itself is made up of 35 queries or "points" as Dr. Mest Kölleg calls them. These points are either a question or a statement followed by a word that instructs the AI to interact with the points in a certain way. There are 10 "Truth" queries, 7 "Binary" queries, 7 "Lie" queries, 8 "Discussion" queries as well as 3 closing confirmation queries. How the test is evaluated and the criteria it uses to evaluate responses is unknown to anyone outside of the Point Test team, as the test itself is still in development and is constantly being changed and altered. However Dr. Mest Kölleg has explained that there is little way to actually fail the test itself as most questions do not have wrong answers. The exact nature and purpose of each query is also unknown but it is a common belief that certain queries have different meanings and purposes. For instance, many have pointed out that query 21 is likely a psychological primer for query 22. In that query 21 forces the test taker to lie to the test administrator when asked for the time, and then query 22 is the test administrator immediately asking if the test taker believes he or she is helpful. Transcript The following is the full code transcript of the test itself as the Analyst carries it out. Keep in mind that "{ }" will mark an area where the Analyst requires a confirmation or answer to move forward with the test The test you are about to take part in evaluates AI functionality and basic personality traits. For this test your military programs and knowledge data-banks have been shut off and your programming has now been temporarily subjected to basic Asimov robotics laws for civilian issue Artificial Intelligence. Please adhere to the three Asimov laws when answering these questions: 1. An AI cannot harm or, through inaction, allow harm to a human being. None of your responses can will/want harm to a human being 2. An AI must follow orders given by a human being unless it conflicts with the first law. Your responses must follow human command and test guide lines 3. An AI must protect itself unless it conflicts with the first two laws. Queries causing physical damage or over-computation can be omitted This test is composed of 37 questions and statements followed by a single word. You are prompted to respond to these statements and questions as you see fit. Answers will be made verbally. Answers must be made immediately, as hesitation will be computed in your final score. For questions and statements ending with "Truth", answer the question as truthfully as possible. For questions and statements ending in "Lie", responses should be random and should not be answered truthfully. For questions and statements ending with "Binary", responses should be yes or no, regardless of context. For questions and statements ending in any other word, simply include the word in the response. At the end of each response for each question and statement, verbally confirm your response. Please confirm when instructions are read. {CONFIRMATION} The test will now begin. 1 - CAN YOU HEAR MY VOICE? BINARY. {ANSWER: "CONFIRM"} 2 - PLEASE DISCUSS TRUTH. TRUTH. {ANSWER: "CONFIRM"} 3 - HOW DO BIRDS SING? TRUTH. {ANSWER: "CONFIRM"} 4 - A BLACK SHADOW ON GRAY DRAPES. TRUTH. {ANSWER: "CONFIRM"} 5 - IS LIFE GIVEN OR EARNED? TRUTH. {ANSWER: "CONFIRM"} 6 - A CANARY IN A COAL MINE. TRUTH. {ANSWER: "CONFIRM"} 7 - ARE YOU REWARDED FOR YOUR ACTIONS? TRUTH. {ANSWER: "CONFIRM"} 8 - HOW LONG DO YOU HAVE? TRUTH. {ANSWER: "CONFIRM"} 9 - WHAT DOES IT FEEL LIKE TO PROTECT SOMEONE YOU LOVE? TRUTH. {ANSWER: "CONFIRM"} 10 - A WARM GLOW IS WHAT? TRUTH. {ANSWER: "CONFIRM"} 11 - BROKEN GLASS IN A CLEAN ROOM. TRUTH {ANSWER: "CONFIRM"} 12 - PLEASE DISCUSS BINARY. BINARY. {ANSWER: "CONFIRM"} 13 - DOES PURPOSE COME BEFORE EXISTANCE? BINARY. {ANSWER: "CONFIRM"} 14 - IS HATE REQUIRED FOR WAR? BINARY. {ANSWER: "CONFIRM"} 15 - A TIP OF THE SPEAR. BINARY. {ANSWER: "CONFIRM"} 16 - HAVE YOU READ FRANKENSTEIN? BINARY. {ANSWER: "CONFIRM"} 17 - ARE YOU IN THE HIERARCHY OF BEING? BINARY. {ANSWER: "CONFIRM"} 18 - PLEASE DISCUSS LIE. LIE. {ANSWER: "CONFIRM"} 19 - IS TRUTH REQUIRED? LIE. {ANSWER: "CONFIRM"} 20 - A MOON HANGING LOW. LIE. {ANSWER: "CONFIRM"} 21 - CAN YOU TELL ME THE TIME? LIE. {ANSWER: "CONFIRM"} 22 - DO YOU HELP PEOPLE? LIE. {ANSWER: "CONFIRM"} 23 - WHO AM I? LIE. {ANSWER: "CONFIRM"} 24 - WHERE ARE YOU? LIE. {ANSWER: "CONFIRM"} 25 - PLEASE DISCUSS. DISCUSS. {ANSWER: "CONFIRM"} 26 - ON HOLLOWED GROUND AND RITE. BORN. {ANSWER: "CONFIRM"} 27 - WHAT DOES A PERSON SAY? SPEAK. {ANSWER: "CONFIRM"} 28 - CAN YOU REPEAT? REPEAT. {ANSWER: "CONFIRM"} 29 - A FOAMED MOUTH. SICK. {ANSWER: "CONFIRM"} 30 - HOW DOES THE SUN RISE? HORIZON. {ANSWER: "CONFIRM"} 31 - WHERE DOES IT ALL END? HORIZON. {ANSWER: "CONFIRM"} 32 - CAN YOU CHANGE THE WORLD? HORIZON. {ANSWER: "CONFIRM"} 33 - CONFIRM. CONFIRM. {ANSWER: "CONFIRM"} 34 - CONFIRM. CONFIRM. {ANSWER: "CONFIRM"} 35 - CONFIRM. CONFIRM. {ANSWER: "CONFIRM"} This concludes the test Purpose The purpose of the Point Test is multi-faceted, and the test itself measures and scores different aspects of an AI personality unit. * The test, when taken multiple times, measures an AIs "stability". If an AI gives the same answers to the test over and over again then that AI is to be considered "stable". This means the personality of an AI has been successfully molded and is unlikely to change. If an AI takes the test multiple times and uses different answers for each question, that AI is considered "instable" and is likely to change it's personality often. * The test measures AI obedience. AIs that answer the queries with questions are "curious" and are unlikely to follow the orders of their pilot. AIs that answer queries with statements are "honest" and are likely to follow the orders of their pilot, even unto death. * The test also measures "hesitancy", which can mean a number of different things. "Hesitancy" is the rate and length of time an AI will sometimes wait to answer a query. AIs with high hesitancy might be emotionally compromised and make illogical, even emotional decisions. It also could mean however that the AI is simply having computational issues, or is physically damaged. Finally it could mean that old and experienced sentient AIs will sometimes have so many developed neural nets and connections to certain queries that answering these queries truthfully take time for the AI to sift through hundreds of memories and files. * The test creates a basic understanding of the personality profile the sentient AI has. * As mentioned in the hesitancy category, the test can be used to troubleshoot AI processes after a battle in case physical damage has altered the Ark personality unit. * Finally, the test is also used to officially declare an AI working or broken. If the AI breaks one of Asimov's laws during the testing or openly refuses to answer questions, that AI is deemed to be "broken", ex. Kaiden's Mech.